*2.4. Genomes Analyzed*

Genome sequences from *T. cruzi* strains were downloaded from TritrypDB [23]: Dm28c 2018 (version 2018-05-30 release 46); Y C6 (version 2019-08-26 release 46); TCC (version 2018-05-30 release 46); CL Brener\_S (version 2015-12-07 release 46); CL Brener\_P (version 2015-12-07 release 46); marinkellei (version 1.0); Brazil A4 (version 2019-08-26 version 46); and Sylvio X10/1 (version 2017-03-18).

#### *2.5. Monte Carlo Test of a 241 nt Repeat*

To test whether the frequency of repeats associated with trans-sialidases was higher than expected by chance, we conducted a Monte-Carlo test [24], in which the total number of repeats found in the CL Brener S genome sequence (334) and in Dm28c strain (1117) were randomly re-inserted in the genome sequence. For each replicate, first a "fake" long and single chromosome was generated by concatenating the 41 chromosomes end-to-end from CL Brener\_S and all 636 contigs of DM28c. Then, the repeats were randomly re-inserted between the first and last nucleotide of this "fake" chromosome. Finally, we determined how many of these re-inserted repeats had a trans-sialidase in their surroundings (either with the repeat falling into it or at 5′ and/or 3′ ). The rationale is that if the original number of repeats were to be inserted from scratch along chromosomes/contigs (for instance, by a natural biological process), then the distribution among the pseudoreplicates (334 draws per peudoreplicate for CL Brener S and 1117 for Dm28c) can be considered a reasonable indicator of the variability of the probability of random insertions near trans-sialidases. If the original value of repeats associated with such genes falls within the 95% highest density interval (HDI) of that distribution of the pseudoreplicates, then the hypothesis of random association with trans-sialidases in the original chromosomes/contigs cannot be discarded; on the contrary, if the original value of trans-sialidase associations is outside the 95% HDI (either higher or lower), then the hypothesis of random association with trans-sialidases can be ruled out at an α = 0.05 level.
